Kinetic studies show that most small proteins (ribonuclease, chymotrypsinogen, cytochrome C, myoglobin, and lysozyme under certain conditions) unfold and refold in two distinct steps; a fast phase with a time constant in the high millisecond range and a slow phase with a time constant in the one second to ten second range near 40 degrees C. It has been generally accepted that these biphasic kinetics prove the existence of high concentrations of states other than the native state and the denatured state. Specific suggestions have been made that proteins may unfold sequentially or that they may become incorrectly folded, but there is not yet any concensus on the particular nature of the structural complexities. Even though kinetic evidence appears to suggest that protein unfolding is not an extremely cooperative process, there is much thermodynamic evidence in the literature which is consistent with the idea that unfolding may be experimentally indistinguishable from a two-state process. It can be shown that there is a model for protein unfolding, not previously considered in the literature, which appears to be capable of rationalizing all existing data. In this model unfolding per se is assumed to occur fast and the redistribution of proline residues between cis and trans isomers is assumed to be the slow step. Data on model compounds show clearly that such a redistribution of isomers would be expected to occur subsequent to unfolding and, furthermore, that the time constant for proline isomerism is to the time constant for the slow phase of protein unfolding. Another desirable feature of the model is that it is easily tested. Experiments utilizing NMR and Stopped-flow spectrophotometry are described which should give a definite answer as to whether or not the slow phase of protein unfolding is rate-limited by proline isomerism.